Hi all,
I've been building a 3-way speaker system for home listening. I've been tossing up between an active and passive system, and recently have been leaning toward a passive x-over design. I've been playing with a design for this in x-sim and have come up with the attached. Is anyone with more knowledge than I able to have a quick glance over this and let me know if there's anything that I should consider?
The drivers I'm using are :
SB Acoustics SB26SFCL38-8 10" Paper Cone Woofer
SB Acoustics SB15NRX2C30-8 Norex Midwoofer
Fountek NeoX1.0 Ribbon Tweeter
I've attached the frequency response which I'm fairly happy with, I don't particularly know how to read the impedence graph and what I should be looking for there, all I've done is try to keep the minimum above 4 Ohms so any pointers here would be great.
The main concern I have is the power dissipation graph which seems to indicate very high power through those resistors. Is there any way to resolve this?
last thing to mention this was constructed from FRD files found for these drivers online, I'll try to take some actual readings once I've loaded the drivers into the cabinets to get a more accurate picture but any help in the meantime would be great!
many thanks
Lawrence
I've been building a 3-way speaker system for home listening. I've been tossing up between an active and passive system, and recently have been leaning toward a passive x-over design. I've been playing with a design for this in x-sim and have come up with the attached. Is anyone with more knowledge than I able to have a quick glance over this and let me know if there's anything that I should consider?
The drivers I'm using are :
SB Acoustics SB26SFCL38-8 10" Paper Cone Woofer
SB Acoustics SB15NRX2C30-8 Norex Midwoofer
Fountek NeoX1.0 Ribbon Tweeter
I've attached the frequency response which I'm fairly happy with, I don't particularly know how to read the impedence graph and what I should be looking for there, all I've done is try to keep the minimum above 4 Ohms so any pointers here would be great.
The main concern I have is the power dissipation graph which seems to indicate very high power through those resistors. Is there any way to resolve this?
last thing to mention this was constructed from FRD files found for these drivers online, I'll try to take some actual readings once I've loaded the drivers into the cabinets to get a more accurate picture but any help in the meantime would be great!
many thanks
Lawrence
Attachments
At a quick glance, I think L4 is a bit too small.
Why is there an R3? Is that something to simulate the amp output impedance?
I don't mess with ribbon tweeters, but I hear that you must be careful. In the sim, try a 6 ohm, and 6 ohm tweeter pad placed after the x-over. Adjust from there.
Why is there an R3? Is that something to simulate the amp output impedance?
I don't mess with ribbon tweeters, but I hear that you must be careful. In the sim, try a 6 ohm, and 6 ohm tweeter pad placed after the x-over. Adjust from there.
R3 is complete non-sense. There has to be no resistor in series with the woofer. Remove it and start again.
Anyway, building a working 3-way passive is something you need years of practical experience, actual measurements of the raw drivers in the cabinet / on the baffle and a good simulation for a start. If you are that far, you need a stash of expensive crossover parts to find the final, working values.
Or short: You have no chance of getting a 3-way right, use it!
A 3-way active is something you could get right, of course only with measuring it. At least not any experimental crossover will need parts for maybe 200$.
Anyway, building a working 3-way passive is something you need years of practical experience, actual measurements of the raw drivers in the cabinet / on the baffle and a good simulation for a start. If you are that far, you need a stash of expensive crossover parts to find the final, working values.
Or short: You have no chance of getting a 3-way right, use it!
A 3-way active is something you could get right, of course only with measuring it. At least not any experimental crossover will need parts for maybe 200$.
You should spend some time learning VituixCAD and then use it instead of x-sim, it will be worth it. As the others mentioned you need to remove the resistor in series with the woofer. When estimating power when using VituixCAD I select pink noise to get reasonable power values for resistors, I don't know if x-sim has that feature. A lot of of people here are going to steer you away from designing your own 3-way, don't ignore them. On the other hand if you are willing to learn what is required for a 3-way, acquire measurement tools and be very patient it will be a rewarding experience. I recommend reading @hifijim posts, he documents very clearly what is required for success.
Get an non-USB microphone, an audio interface and a microphone stand and start measuring speakers before buying parts or cutting wood. Once you can reliably take measurements buy some drivers and a cheap DSP and build a test enclosure. At this pint you will be able to get the measurements required to begin simulations and use the DSP to start figuring out how to make the drivers work together. Lastly, don't make the mistakes I made by buying crossover parts based solely on simulations using manufactures data or you will end up with lots of spare parts.
The drivers you listed might work but the 5 inch mid-woofer crossed to the ribbon might not be optimal. I built something similar with and 8 inch woofer, a 4 inch mid-range (not a mid-woofer) and the same tweeter. https://bellarossafabrica.net/brf-38b
Get an non-USB microphone, an audio interface and a microphone stand and start measuring speakers before buying parts or cutting wood. Once you can reliably take measurements buy some drivers and a cheap DSP and build a test enclosure. At this pint you will be able to get the measurements required to begin simulations and use the DSP to start figuring out how to make the drivers work together. Lastly, don't make the mistakes I made by buying crossover parts based solely on simulations using manufactures data or you will end up with lots of spare parts.
The drivers you listed might work but the 5 inch mid-woofer crossed to the ribbon might not be optimal. I built something similar with and 8 inch woofer, a 4 inch mid-range (not a mid-woofer) and the same tweeter. https://bellarossafabrica.net/brf-38b
I don't believe it's time to discourage you from a 3-way since there are so many other things to talk about first. I won't make a list, but as for what's in front of us...
You're not showing phase. Here's a thread on Xsim plot window setup.
https://www.diyaudio.com/community/threads/xsim-configuring-the-plot-window.406106/
If the drivers aren't in the cabinets then there will be differences enough when you measure them.
Usually, but occasionally there can be. I'd remove it in this case though as all three have a resistor, and it is recommended to make one the base for sensitivity and adjust the other two.
What the others said:
1. remove R3
2. replace R1 & R2 with Lpads, after the xover, and before the driver
3. the reason to use Lpads instead of series resistors is the impedance that a particular section of the xover sees stays the same, with a series resistor you'll have to re-calculate the values of C's and L's
4. your woofer will define the overall sensitivity, use its sensitivity to calculate how many dBs to cut from the other drivers with the Lpads
5. as Allen above said, check phase, the normal practice is the woofer is connected with positive polarity, the other drivers can be either/or, the phase and freq. response will tell you which way to go. Your arrangement is also valid but needs to be verified.
6. cut the woofer at the Baffle Step, which means you need to know the size of your baffle.
and so on...
1. remove R3
2. replace R1 & R2 with Lpads, after the xover, and before the driver
3. the reason to use Lpads instead of series resistors is the impedance that a particular section of the xover sees stays the same, with a series resistor you'll have to re-calculate the values of C's and L's
4. your woofer will define the overall sensitivity, use its sensitivity to calculate how many dBs to cut from the other drivers with the Lpads
5. as Allen above said, check phase, the normal practice is the woofer is connected with positive polarity, the other drivers can be either/or, the phase and freq. response will tell you which way to go. Your arrangement is also valid but needs to be verified.
6. cut the woofer at the Baffle Step, which means you need to know the size of your baffle.
and so on...
Thanks all, R3 was to attenuate a hump around 50Hz I'd added in as a sort of afterthought - I take the feedback on board though I should be using one driver to define sensitivity - thanks.
I'll put the drivers into the cabinet and start taking proper measurements in order to do another sim from there.
I'm well aware of how difficult everyone says that 3 way passive systems are to design. I've built sound systems in the past although always with active crossovers so I'm keen to learn something new here.
thanks to everyone for the input though - I appreciate all the feedback! looks like I'll have to get vituix out again and have a play with that...
I'll put the drivers into the cabinet and start taking proper measurements in order to do another sim from there.
I'm well aware of how difficult everyone says that 3 way passive systems are to design. I've built sound systems in the past although always with active crossovers so I'm keen to learn something new here.
thanks to everyone for the input though - I appreciate all the feedback! looks like I'll have to get vituix out again and have a play with that...
I'm careful when I hear this because I've seen the reasons given change over the years. For example it once used to be about filter interaction on the mid, but then simulators became readily available.. so I'd want to hear someone say why before coming to a conclusion. The fact is that some of this knowledge is well needed for good two way design, too 😉
As I see it, polar measurements tackle the most obvious questions hanging over a 3-way design since they set the potential between some of the more typical 2 and 3 way. In a sense it can be fairly easy if you have the right information in front of you and are experienced enough to use it.
On principle, you lower the levels of your most sensitive drivers to the least sensitive one, after equalization. Since one of the 3 drivers is always the least efficient, it gets no padding per se. This is often the woofer. You set your low pass filter, and set your baffle step and then the mid and tweeter are brought down in line to it.
It is great that FRD and ZMA are widely available these days.
And at least your set on the drivers. Cause we often see brain storms posted for 5 different drivers.
Then most the data is in sim with wrong data anyways.
Long story short factory FRD is in ideal Half space.
Needs to be converted to full space and show half space loading of the exact baffle size.
Then positions of the drivers needs to be set to exact positions so the phase is correct.
So again at least your set on the drivers and time and effort can be put into just that.
Getting accurate frequency and phase response to accurately sim.
So almost always it leads to Virtuix Cad which will also generate off axis data.
Get tired of seeing Xsim presentations with no phase or off axis simulation.
Designing 0 degree on Axis with no vertical off axis. Doesnt teach people much
about phase and center to center placement or drivers real world.
No offense it is just a long long learning curve.
Virtuix Cad has numerous filter design aids and will give you off axis data.
And the FRD files generated will show the big difference with on baffle response.
So ideal inductor values are not even chosen by many experienced builders.
And impedance compensation is often misunderstood.
So not only do we need accurate driver position and phase.
The user library in Virtuix Cad will give you examples of correct padding, impedance compensation
and fully adjustable 3 ways crossovers. Which can fast track years of learning the hard way.
You will deal with less errors and wild impedance curves, much easier.
Since a " 8 ohm" driver wont always use generic values that Xsim will leave you guessing at wildly
based on generic theory.
And at least your set on the drivers. Cause we often see brain storms posted for 5 different drivers.
Then most the data is in sim with wrong data anyways.
Long story short factory FRD is in ideal Half space.
Needs to be converted to full space and show half space loading of the exact baffle size.
Then positions of the drivers needs to be set to exact positions so the phase is correct.
So again at least your set on the drivers and time and effort can be put into just that.
Getting accurate frequency and phase response to accurately sim.
So almost always it leads to Virtuix Cad which will also generate off axis data.
Get tired of seeing Xsim presentations with no phase or off axis simulation.
Designing 0 degree on Axis with no vertical off axis. Doesnt teach people much
about phase and center to center placement or drivers real world.
No offense it is just a long long learning curve.
Virtuix Cad has numerous filter design aids and will give you off axis data.
And the FRD files generated will show the big difference with on baffle response.
So ideal inductor values are not even chosen by many experienced builders.
And impedance compensation is often misunderstood.
So not only do we need accurate driver position and phase.
The user library in Virtuix Cad will give you examples of correct padding, impedance compensation
and fully adjustable 3 ways crossovers. Which can fast track years of learning the hard way.
You will deal with less errors and wild impedance curves, much easier.
Since a " 8 ohm" driver wont always use generic values that Xsim will leave you guessing at wildly
based on generic theory.
Many thanks all , this has been illuminating as any post on here always is. I’ll spend some time measuring and gaining real frd files before running them through vituix . Next time I come back I’ll make sure to include phase diagrams too.
When taking measurements would you recommend both on an off axis then? Should 3 off axis (30/60/90) be sufficient to gain an idea of off axis responses or would you suggest more?
When taking measurements would you recommend both on an off axis then? Should 3 off axis (30/60/90) be sufficient to gain an idea of off axis responses or would you suggest more?
I've never designed a 3-way, but have had success with several 2-way passive designs. I agree with all the advice I've read here, especially the advice that you measure your drivers in the speaker cabinet.
Have you considered bi-amping with a passive X-over between midrange and tweeter? That would simplify measurements because you wouldn't need to measure lower frequencies. Also: inductors and capacitors get more expensive as the frequencies they're filtering get lower. For example: That 3.0mH inductor costs from $15.98 to $45.75 at Parts express (depending on brand and AWG)
To answer your last question: Yes, I recommend measuring off-axis.
Have you considered bi-amping with a passive X-over between midrange and tweeter? That would simplify measurements because you wouldn't need to measure lower frequencies. Also: inductors and capacitors get more expensive as the frequencies they're filtering get lower. For example: That 3.0mH inductor costs from $15.98 to $45.75 at Parts express (depending on brand and AWG)
To answer your last question: Yes, I recommend measuring off-axis.
The off-axis measurements help you gauge the right crossover points. For the mid to tweeter handoff for instance, a higher crossover point is desirable to improve the dynamic range and reduce the power and distortion on the tweeter, but it also risks the mid beaming and reducing the off-axis response.
The same kind of thing happens at the woofer/mid handoff. I wrote about this in detail here.
Having said all that, while many here have the time, energy and space to run multiple off-axis tests I really just do 45 degrees. That's good enough to show me where the crossover points should be and then I build based on that. I don't feel the need to do a 360 degree simulation really and I'm usually pleased with the results.
Coincidentally I was playing with the same woofer and midrange (tweeter is sb26stac). The baffle is 1100x300mm, woofer cabinet is 65L sealed full stuffed, mid cabinet is 4L full stuffed.
Last edited:
Great, @shadowplay62 , but you think this SB15NRX2 really wants to run up to 3k? I had NRX1 and wouldn't get it again; lows "alright" but middle midrange and upper midrange sound not to my taste. Maybe NRX2 is improved.
The most important tool to have is an analog calibrated microphone. It's worth the money.
The FRD files you get from manufacturer don't include the phase difference caused by geometrical differences you'll have in your own box/baffle.
With the calibrated mic and using REW software, if you make timed measurements (very important), you get your own FRD produced by your speakers in your box. So any geometrical characteristics will be embedded in the FRD file.
The exact distance from each speaker to the mic will be captured and this makes a lot of difference when combining all speakers in the crossover simulation software.
Regarding power dissipation, XSim probably considers a pure and constant sine wave with maximum power in each frequency.
The simulation is correct, but it's not what happens when you play music.
You can build your speaker to support this scenario, but it will be non sense and very expensive. No point at all.
You must consider some deratings:
1. Music doesn't have equal power in all frequencies. More power on low frequencies and less power on higher frequencies.
A 100W amp playing music will typically delivers let's say 80% to woofer, 15% to mid and 5% to tweeter - just an example - it varies a lot with music and how do you listen to it. As mentioned, some people use pink noise.
2. Music is dynamic, so you have to consider a kind of duty cycle proportion.
Example: 100W playing a single drum kick at maximum power, will have a duty cycle of let's say 20% (20% on and 80% off).
So the average power delivered to the woofer in this scenario will be just 20W. Same will apply to the resistors involved in the crossover.
That's why the tweeter, for example, is normally rated as "system power" and not exactly it's own average power.
A 5W tweeter (actual average power that can be dissapated) might be rated for a "system power" of 100W.
We can easily fry a tweeter if we apply a pure 10kHz tone with 100W to a loudspeaker rated for 100W.
But we can apply 100W in the woofer using let's say 80Hz and nothing happens.
This apply to all other components in the loudspeaker (speakers, resistors, inductors etc).
Wow thanks so much! that looks incredibly complicated and not something I imagine I'd have been able to design myself but might give me a good basis once I do get some measurements of my own drivers in their boxes!
thanks also to everyone else for the detailed inputs!
@shadowplay62 How did you get VituixCAD to display the three dashed target curves?
@ljettoo and @ByronInPortland The XO isn't that complicated, it is just laid out a little oddly. The woofer circuit is 2nd order with two caps to get the wanted value and a notch filter. The mid is also a 2nd order with a notch, and an LPAD split between the amplifier side and the driver side. The tweeter is very normal, a 2nd ordre and an LPad.
@ljettoo and @ByronInPortland The XO isn't that complicated, it is just laid out a little oddly. The woofer circuit is 2nd order with two caps to get the wanted value and a notch filter. The mid is also a 2nd order with a notch, and an LPAD split between the amplifier side and the driver side. The tweeter is very normal, a 2nd ordre and an LPad.